Horne and Lowe (1996) suggest that learning a common name for several
arbitrary, or disparate, stimuli may establish a category relation between these
stimuli. The present studies explore in 3- to 4.5-year-old children whether categories at different levels can be established less directly by training intraverbal relations between lower-level names and potential higher-level category names.
In Study 1a, the children were trained to tact (see animal - say name) eight
different "alien" animals randomly allocated to four two-member common name
categories. When tested, all 11 children who were tested showed the corresponding untrained listener behaviour (hear /name/ - select animal).
Study 1b linked these lower-level names (hib, feb, tor, and lup) to potential
higher-level names (zaag and noom) in an echoic and intraverbal game. Following this, 5 out of 8 children showed correct listener behaviour at the higher name level in Leg 1 of the study, and 3 out of 5 in Leg 2.
In Study 1c, two gestures were trained, one to a zaag, one to a noom. One of
the 3 children showed untrained transfer of function to all other aliens, via lowerand higher-level names. Another child showed partial transfer.
In Study 1d, animal cries were trained to two different aliens, again to one
zaag and one noom. The one remaining child showed partial transfer.
Study 1e found that a more general verbal prompt than that used in Studies 1c
and 1d elicited the appropriate gestures and animal cries, but never both on the same trial.
Studies 1f and 1g found listener behaviour to the gestures and animal cries to
be in place.
When tested in Study 1h, the child correctly sorted all stimuli into lower- and
higher-level categories.
Studies 2a-h replicated these studies, but employed pre-training with familiar
stimuli before the alien Studies 2a, 2b, 2c, and 2h. Alien tact training required fewer trials than in Studies 1a-1h, and 3 more children (7/8, and one more in retests) showed correct listener behaviour at the higher name level in Leg 1, while 3 out of 7 passed for Leg 2. However, transfer of function results were as in Studies 1, while in Study 2h the child sorted stimuli correctly into lower-, but not higher-level categories.
The data are consistent with a Skinnerian account of verbal behaviour.